When you are feeling down it is comforting to listen to sad music as it matches your mood. Sometimes the music can make you feel less alone or it feels like someone else relates to your problems. The corpus I chose is a playlist full of sad songs. A very good playlist of a friend of mine called “Sad *itch Hours” is a playlist with 203 sad songs. I find it interesting that there are also different playlists by Spotify that focus on sad music, but also differentiate in, for example, heartbreak or just having a bad day. I am interested to learn what makes a playlist a sad playlist? To do this, I choose to include the Spotify playlist “Life Sucks” in my corpus, which has 200 sad songs. In contrast to this I would also like to add playlists focussed on happy moods to my corpus, to look at the differences between happy and sad music. This will include “Happy Favorites” and “Wake Up Happy”, which are both playlists made by Spotify and containing 150 songs each. All playlist included in this corpus contain songs of various genres and artists, not focussing on one specific genre.
The natural groups I would like to investigate in the sad playlist is artists and valence of songs. It is interesting to see which artist has the most songs in the sad playlists and whether an artist might be more diverse by having both happy and sad songs. Futhermore, focussing on the difference between the spotify generated playlist and the one of my friend, to see if there is anything that stands out. Additionally, I would like to see whether there are factors that influence valence, for example, danceability or speechiness, and to compare this with happy and sad music.
I believe the tracks in my corpus are very representative as the playlist “Sad *itch Hours” has a very diverse group of artists and therefore also genres. However, it does miss genres like country or metal sad music, as most songs are slow hip-hop, r&b or pop songs. The playlist generated by Spotify of sad and happy music are also very broad, but do focus on music that is more recent, and don’t include many old songs. I chose playlists that are not focussed on one specific genre to make a better comparison.
Typical tracks in my corpus for sad songs are Don’t Speak by No Doubt, songs by Billie Eilish, Lana del Rey, or Adele. Looking at the happy playlist by Spotify, it varies a lot from feel good songs to real party anthems.
To start off I wanted to see what the level of valence is according to Spotify in the two sad playlists. I used the histogram to focus on only one feature. According to Spotify, valence describes the musical positiveness of a song from 0 to 1, meaning more negative and more positive respectively. This figure shows that the playlist made by Spotify (“Life Sucks”) is slightly more positive than the one my friend made, as it contains more songs with a higher valence value.
Then I wanted to look at the differences between the happy and sad playlists, and look at features that might influence valence. Therefore, this next visualisation includes valence, danceablity and energy. Danceability rates how suitible a song is to dance to (1.0 being very danceable) and energy represents the intensity or activity, which is calculated with loudness, timbre, dynamic range, etc. (1.0 being very energetic). In this visualisation it is seen that the distribution of valence is broad in both happy and sad playlists, but there is also a division showing that overall happy music has higher valence than sad music, which was also expected. It also shows that positive music has a higher danceability, and the energy of positive music is mostly above the 0.50. Whereas only the blue dots (sad songs) have an energy of 0.25.
This chromagram resembles the song Another Love by Tom Odell. A song that is more than 10 years old, but gained popularity in 2022 due to TikTok. It is a song that is part of the sad songs corpus. What I find interesting about this chromagram is how it starts off with more indiviual notes, but around 100 seconds in, there is a change. This is when the choir starts singing in the song and the tempo speeds up. In the chromagram it looks like the notes are more blurry and blend into each other.